for nonsense COLSPANs at the edge of a table. Fixed some bugs this uncovered
where the cell was being asked for its colspan attribute rather than the table
being asked for the cell's effective col span.
* fixed more margin code. we're inching towards correctness.
* the width contributed by a colspanning cell is now proportionately distributed to the columns based on the columns "effective width"
* added logic for handling illegal COLS attribute on a table (like COLS=4 in a 3-column table). See nsTableFrame::GetEffectiveCOLSAttribute()
* added logic for handling illegal COLSPAN attribute on a cell (like COLSPAN=4 in a 3-column table). See nsTableFrame::GetEffectiveColSpan(). This is wrong, and will get reworked soon.
* added the ability for cell width attributes to effect column width like Nav4.
added backwards compatibility for percent width tables inside of auto width tables (see http://webreview.com/wr/pub warning: you'll have to look at a local copy until
Troy/Vidur/somebody fixes an image problem.)
this work exposed a few bugs and slow spots, which have been fixed
for the aol page, I added some additional backwards compatibility code
to proportionately distribute width when a fixed-width cell has colspans
1. cellmap couldn't properly delete CellData because definition was unavailable
2. optimized table cells made taller wouldn't shrink when they should because we were not
remembering the previous desired height of the cell.
3. rows were placing cells on the left edge, and not adding in the left margin.
from the prior available width. Some real-world test cases sped up
1-2 orders of magnitude for resize reflow. Initial reflow is unchanged because
we still need to get pass1 metrics on all content.
fixed nsCSSBlockFrame.cpp to take margins into account when computing parent's available width
fixed distribution of excess space to table cells when table is bigger than the content it holds.
not quite finished yet, but better.
of subsequent cells in reflow pass 1
we correctly distribute extra space in a table proportionately.
We try to give the extra space to auto-width columns first,
and if there are none we distribute it to all columns.
The percentage is based on the cell content's desired size.